A method and system for detecting stator insulation weakening in permanent magnet motors based on leakage current.

Abstract

This invention belongs to the field of motor insulation testing, specifically relating to a method and system for detecting stator insulation weakening in permanent magnet motors based on leakage current. The method includes: calibrating the characteristic frequency points of the permanent magnet motor; calculating the energy at the characteristic frequency points based on the narrow bandwidth and the characteristic frequency points; collecting multiple operating points and expected energies of the motor under normal operating conditions; constructing a benchmark database based on the benchmark energy, operating points, and expected energies; training an energy prediction model using the benchmark database to obtain an optimal energy prediction model; inputting the actual operating conditions into the optimal energy prediction model to predict the predicted energy corresponding to each characteristic frequency point; calculating the calculated energy based on the actual operating conditions and each characteristic frequency point; and obtaining the detection result of stator insulation weakening in the permanent magnet motor based on the calculated energy and the energy difference. This application has the effect of improving detection sensitivity and increasing the accuracy of insulation detection.

Description

Technical Field

[0001] This invention belongs to the field of motor insulation testing, specifically relating to a method and system for detecting weakened stator insulation of permanent magnet motors based on leakage current. Background Technology

[0002] Permanent magnet synchronous motors (PMSMs), with their high power density, high efficiency, and excellent dynamic response, have become the core power unit of rail transit traction systems, playing a crucial role, especially in high-speed trains. Their high power-to-weight ratio and lightweight characteristics significantly improve train traction efficiency and energy utilization, while meeting the stringent requirements of high-speed operation for system stability and energy consumption control. Their operational reliability directly determines the traction safety, life-cycle energy efficiency, and service life of high-speed trains. However, the stator winding insulation system, as a critical barrier to the electrical integrity of the motor, faces severe challenges under complex and demanding operating conditions.

[0003] Statistical studies show that insulation degradation accounts for over 40% of motor failures, making it a major precursor to catastrophic faults (such as phase-to-phase short circuits and grounding faults). The gradual weakening process of stator insulation is particularly insidious, with subtle early characteristics that are difficult to identify effectively using traditional methods. Notably, under high-frequency PWM inverter power supply, especially with the use of wide-bandgap semiconductor devices (SiC / GaN), the dramatic increase in switching speed (dV / dt reaching over 50kV / μs) leads to extremely high transient voltage stress at the winding ends and within the slots, exacerbating partial discharge activity within the insulation medium and significantly accelerating the electro-thermal-mechanical aging process of the insulation material. Research data indicates that when the insulation resistance drops to 30% of its initial value, the partial discharge initiation voltage may have decreased by more than 50%. At this point, the high-frequency component of the leakage current exhibits a noticeable characteristic increase, but the overall performance has not yet shown significant degradation. If warnings are not issued in this early stage, insulation weakening will continue to develop, potentially leading to a chain reaction of arc discharge, winding burnout, and even irreversible demagnetization of permanent magnets, causing significant safety risks and economic losses.

[0004] In recent years, leakage current-based monitoring methods have shown unique potential in the field of insulation diagnostics. Both theoretical analysis and experimental verification have demonstrated that the overall dielectric properties of the stator insulation system (such as equivalent capacitance and resistance) and its internal partial discharge activity are directly reflected in the amplitude, waveform, and harmonic components of the leakage current in the motor common-mode circuit or protective grounding wire. In particular, specific spectral components in the leakage current signal related to the PWM switching frequency and its harmonics are highly sensitive to increases in insulation dielectric loss, enhanced interface polarization effects, and increased partial discharge intensity. However, traditional online leakage current-based detection methods typically focus on identifying extreme points (such as peak values ​​and RMS values) of the spectral amplitude in specific frequency bands (such as the area near the switching frequency and its harmonics). This method faces the following challenges in practice: Weak anti-interference capability: The leakage current signal spectrum is easily polluted by motor drive control strategies (such as overmodulation and dead zone compensation) and external strong electromagnetic interference (EMI), resulting in drastic fluctuations in amplitude at key frequency points and low signal-to-noise ratio; Insufficient feature sensitivity: In the early stage of the performance degradation of insulating materials, the change amplitude of their state characterization features is small and the spectrum distribution is wide, making it difficult to extract statistically significant and stable fault feature fingerprints from the amplitude response of a single frequency point, thus limiting the sensitivity and reliability of fault diagnosis based on frequency domain analysis. Summary of the Invention

[0005] The technical problem this invention aims to solve is to provide a method and system for detecting stator insulation weakening in permanent magnet motors based on leakage current. By deeply mining the energy distribution evolution of multiple specific frequency bands in the leakage current signal that are strongly correlated with the insulation state, a highly discriminative fault quantification index is constructed. This method can accurately capture early and subtle insulation degradation characteristics under strong noise backgrounds, significantly improving detection sensitivity and accuracy.

[0006] A method for detecting stator insulation weakening in permanent magnet motors based on leakage current, comprising: The characteristic frequency points of the permanent magnet motor are calibrated, and there are 3 characteristic frequency points; Set a narrow bandwidth for each of the aforementioned characteristic frequency points; Based on the narrow bandwidth and the characteristic frequency points, the characteristic frequency point energy corresponding to each characteristic frequency point is calculated, and the characteristic frequency point energy is used as the reference energy. Collect multiple operating points of the motor under normal operating conditions and the expected energy corresponding to each operating point; A benchmark database is formed based on the benchmark energy, operating point, and the expected energy corresponding to the operating point. The energy prediction model is trained using the benchmark database to obtain the optimal parameters, and the energy prediction model is optimized based on the optimal parameters to obtain the optimal energy prediction model. The actual working conditions are input into the optimal energy prediction model to predict the predicted energy corresponding to each characteristic frequency point. The predicted energy includes the first predicted energy, the second predicted energy, and the third predicted energy. Based on the actual working conditions and each characteristic frequency point, the first calculated energy, the second calculated energy, and the third calculated energy are calculated. Based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy, the first energy difference, the second energy difference, and the third energy difference are calculated. Based on the first energy difference, the second energy difference, and the third energy difference, the detection results of the permanent magnet motor stator insulation weakening are obtained.

[0007] Optionally, the step of calculating the characteristic frequency energy corresponding to each characteristic frequency point based on the narrow bandwidth and characteristic frequency points, and using the characteristic frequency energy as a reference energy, includes: Set the sampling frequency; Based on the narrow bandwidth and the characteristic frequency point, the upper cutoff frequency and the lower cutoff frequency are obtained; Obtain the original signal when calibrating the characteristic frequency point of the permanent magnet motor; Based on the original signal, the upper cutoff frequency, the lower cutoff frequency, and the sampling frequency, the characteristic frequency energy corresponding to each characteristic frequency point is calculated, and the characteristic frequency energy is used as the reference energy.

[0008] Optionally, each operating point includes speed, torque, and temperature.

[0009] Optionally, the step of training the energy prediction model using the benchmark database to obtain optimal parameters, and optimizing the energy prediction model based on the optimal parameters to obtain the optimal energy prediction model, includes: Each set of training data in the benchmark database is input into the energy prediction model, and the optimal parameters are obtained by the least squares optimization algorithm. Each set of training parameters includes a working point, the benchmark energy corresponding to each feature frequency point, and the expected energy corresponding to each working point. The energy prediction model is optimized based on the optimal parameters to obtain the optimal energy prediction model.

[0010] Optionally, the step of calculating the first energy difference, the second energy difference, and the third energy difference based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy includes: The difference between the first predicted energy and the first calculated energy is obtained to get the first calculated difference value. The absolute value of the first calculated difference value is then obtained to get the first energy difference value. The difference between the second predicted energy and the second calculated energy is obtained to get the second calculated difference value. The absolute value of the second calculated difference value is then obtained to get the second energy difference value. The difference between the third predicted energy and the third calculated energy is used to obtain the third calculated difference value. The absolute value of the third calculated difference value is then used to obtain the third energy difference value.

[0011] Optionally, obtaining the permanent magnet motor stator insulation weakening detection result based on the first energy difference, the second energy difference, and the third energy difference includes: Set a first threshold, a second threshold, a third threshold, a fourth threshold, and a fifth threshold; If the first energy difference is greater than the first threshold and the second and third energy differences are both less than the second threshold, then the result of the permanent magnet motor stator insulation weakening detection is early deterioration. If the second energy difference is greater than the third threshold, the third energy difference is less than the fourth threshold, and the first energy difference continues to rise, then the result of the permanent magnet motor stator insulation weakening test is mid-term deterioration. If the third energy difference is greater than the fifth threshold or the rate of change of adjacent third energy differences is greater than the preset rate of change, then the result of the permanent magnet motor stator insulation weakening detection is late-stage deterioration.

[0012] Optionally, the first threshold is less than the third threshold, and the third threshold is less than the fifth threshold.

[0013] A permanent magnet motor stator insulation weakening detection system based on leakage current, comprising: A calibration module is used to calibrate the characteristic frequency points of the permanent magnet motor, wherein there are three characteristic frequency points; The setting module is used to set the narrow bandwidth for each of the characteristic frequency points; The first calculation module is used to calculate the characteristic frequency energy corresponding to each characteristic frequency point based on the narrow bandwidth and characteristic frequency points, and use the characteristic frequency energy as the reference energy. The data acquisition module is used to collect multiple operating points of the motor when it is operating within the normal operating range, as well as the expected energy corresponding to each operating point. The database construction module is used to form a benchmark database based on the benchmark energy, the operating point, and the expected energy corresponding to the operating point. The optimization module is used to train the energy prediction model using the benchmark database, obtain the optimal parameters, and optimize the energy prediction model based on the optimal parameters to obtain the optimal energy prediction model. The second calculation module is used to input the actual working conditions into the optimal energy prediction model and predict the predicted energy corresponding to each characteristic frequency point. The predicted energy includes the first predicted energy, the second predicted energy, and the third predicted energy. The third calculation module is used to calculate the first calculation energy, the second calculation energy, and the third calculation energy based on the actual working conditions and each characteristic frequency point. The fourth calculation module is used to calculate the first energy difference, the second energy difference, and the third energy difference based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy. The detection module is used to obtain the detection result of the weakening of the stator insulation of the permanent magnet motor based on the first energy difference, the second energy difference, and the third energy difference.

[0014] A terminal device includes a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, it employs a method for detecting weakened stator insulation of a permanent magnet motor based on leakage current.

[0015] A computer-readable storage medium storing a computer program, which, when loaded and executed by a processor, employs a method for detecting weakened stator insulation of a permanent magnet motor based on leakage current.

[0016] The beneficial effects of this invention are: 1. Targeted Energy Extraction: By pre-calibrating to determine the motor's specific resonant frequency, narrowband energy is extracted within the characteristic frequency band, significantly suppressing broadband noise interference and resulting in a significant improvement in characteristic sensitivity; 2. Dynamic Decoupling Architecture: A quantitative mapping model of energy and operating parameters is established to effectively eliminate characteristic baseline drift caused by operating parameters such as speed, load, and temperature, significantly reducing the risk of misjudgment. This solution can achieve highly sensitive identification of minute changes in insulation state, providing reliable insulation safety protection for high-end permanent magnet motors.

[0017] 3. Multi-parameter collaborative evolution hierarchical decision-making: By verifying the logical sequence and state consistency of ΔE1, ΔE2, and ΔE3 in the insulation degradation chain, accurate identification and forward warning of fault stages are achieved, fundamentally avoiding the drawbacks of false alarms and missed alarms in the traditional single threshold method. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the topology of the permanent magnet synchronous motor system of the present invention; Figure 2 The time-domain waveform of the motor leakage current signal is shown in the figure below. Figure 3 This is a framework diagram of a method for detecting stator insulation weakening of a permanent magnet motor based on leakage current according to the present invention. Figure 4 This is a single leakage current spectrum diagram of the present invention; Figure 5 This is a diagram illustrating the hierarchical early warning strategy of the present invention. Detailed Implementation

[0019] Due to its distributed parameter characteristics and high-frequency response, permanent magnet synchronous motor systems form complex impedance networks containing multiple coupling paths. This network encompasses not only the parasitic parameters of the stator windings (resistance and inductance) but also the distributed impedance between the equipment and the grounding system. Its topology is as follows: Figure 1 As shown, when the IGBT device performs high-frequency switching operation, the bus voltage is directly applied to the motor windings through parasitic capacitance, causing transient high-frequency oscillations. In this context, if the inter-turn insulation of the windings deteriorates, key parameters in the equivalent network (especially inter-turn capacitance Cp and inter-turn resistance Rp) will shift, thereby altering the characteristic parameters of the leakage current oscillation waveform. Based on the equivalent circuit model analysis, when the IGBT device switches, the leakage current signal obtained by a three-phase through-cell leakage current sensor installed at the motor input terminal can effectively extract characteristic quantities reflecting changes in inter-turn impedance, thus enabling a quantitative assessment of the motor insulation condition deterioration.

[0020] At the instant the PWM inverter switches, the rate of voltage change (dV / dt) applied to the motor windings reaches its highest level. At this moment, the electrical stress excitation on the insulation system is strongest, generating the most significant transient resonant current response containing rich insulation state information. This current signal, directly excited by the switching transient, has more significant representativeness and higher insulation diagnostic value compared to steady-state operating signals or randomly triggered signals. Therefore, when the detection system accurately captures such a switching transient event, it immediately triggers a high-precision leakage current acquisition channel to synchronously capture the motor leakage current signal (e.g., ...) at a high sampling rate. Figure 2 (The time-domain waveform is shown). The capture operation is strictly limited to a preset time window surrounding the transient event. The core design of this window is to ensure complete coverage of the entire process of high-frequency leakage current oscillation generation and decay under PWM switching excitation. Specifically, the time window structure includes: a leading-edge buffer at the beginning, which ensures complete and unobstructed capture of the starting point of the current response; and the main acquisition area immediately following it. The duration of the main acquisition area is a key design parameter, and its design goal is to fully cover the complete decay process of the main resonant current caused by this PWM switching excitation under typical insulation conditions for this type of motor. To accommodate the differences in electrical characteristics of different motor types (such as winding distributed capacitance, inductance, etc.), the specific duration of this window can be adjusted accordingly.

[0021] This strategy, based on switching transient events, fundamentally transforms / upgrades the broadband noise background problem, which is difficult to avoid in traditional continuous sampling modes, into a focused analysis of specific transient oscillation waveforms, significantly improving the quality and signal-to-noise ratio of the target signal. Its theoretical basis lies in the clear physical coupling mechanism between the switching dynamic behavior of the PWM inverter and the high-frequency response exhibited by the motor leakage current. By accurately identifying and analyzing the key frequency domain characteristics of this transient response, a direct and effective channel for extracting insulation state characteristics can be constructed, ultimately achieving non-invasive quantitative characterization and monitoring of the motor insulation degradation process.

[0022] A method for detecting stator insulation weakening in permanent magnet motors based on leakage current, such as... Figure 3 As shown, it includes: S1. Calibrate the characteristic frequency points of the permanent magnet motor. There are 3 characteristic frequency points. Specifically, three characteristic frequency points are predetermined through offline calibration. The calibration process is performed under reference insulation conditions to obtain stable winding electromagnetic response characteristics. By applying an excitation signal within a predetermined frequency range to the motor stator winding and analyzing its spectral characteristics, three resonant frequency points strongly correlated with specific insulation defects are identified: ① Interlayer capacitance sensitive frequency ( ): The stator winding inter-turn insulation can be equivalent to a distributed RC network, which, together with the slot leakage inductance, forms a series resonant circuit. Insulation degradation (moisture absorption, thermal aging leading to polymer degradation) can potentially increase interlayer capacitance. When interlayer capacitance increases, the resonant frequency shifts to lower frequencies. ,in Interlayer distributed capacitance, (With a narrow bandwidth), while the resonant energy is significantly enhanced due to impedance matching optimization. The energy value extracted from the neighborhood essentially quantifies the evolution of the capacitor's dielectric properties, with the first reference energy E1 ( The enhancement of the corresponding reference energy is not only because The offset, and the degradation may also cause the quality factor Q of the resonant circuit to increase (loss resistance decreases), and the energy to become more concentrated.

[0023] ② Characteristic frequency points of partial discharge ( ): Micrometer-sized air gaps or cracks inside an insulating medium ionize under a strong electric field, generating nanosecond-level partial discharge pulses. These pulses have an extremely wide spectrum, but their energy is concentrated at a specific frequency. Peak values ​​are formed. During the insulation degradation process, air gap defects increase in number and size, and discharge activity intensifies, manifesting as... Narrowband energy peak E2 ( A significant increase in the corresponding baseline energy (E2). E2 directly maps the density of micro-defects inside the insulation and the activity of partial discharge, and is an important medium-term indicator of accelerated insulation degradation.

[0024] ③Characteristic frequency points of carbonization noise ( ): Severe insulation degradation leads to localized carbonization, forming millimeter-scale conductive channels. These channels, under the excitation of the high-frequency electromagnetic field generated by the switching dV / dt, act like miniature antennas embedded in windings, radiating broadband electromagnetic noise. Through offline calibration or theoretical analysis, a specific frequency point that can effectively characterize this noise energy rise needs to be determined. The energy E3 measured in the narrow band near this point ( The step increase in the corresponding baseline energy is essentially a direct measure of the enhanced electromagnetic radiation power in the carbonized region, indicating that the insulation is about to or has already experienced a penetrating failure.

[0025] Three characteristic frequency points ( , , Accurate calibration of the insulation is a prerequisite for achieving high-sensitivity diagnosis using this method. The calibration process is not simply about finding spectral peaks, but rather a process of establishing a "motor insulation feature library" based on a reference insulation state through systematic excitation and refined signal processing. The specific calibration procedure is as follows: I. Systematic Incentives and Data Collection: The experiment was conducted on a test bench when the motor was in a reference insulation condition (such as brand new or confirmed to be healthy).

[0026] Using a programmable power supply or drive controller, simulate the switching behavior of a real PWM inverter by applying a series of step voltage pulses with high dV / dt. To ensure a sufficient frequency response, the pulse rise time should cover the most severe conditions the motor might encounter during actual operation (e.g., nanoseconds). Simultaneously acquire the transient response signal of the leakage current generated by this excitation. To ensure signal integrity, the sampling rate must satisfy the Nyquist sampling theorem, typically set to the highest preset analysis frequency. More than 2.5 times that.

[0027] II. High-resolution spectrum estimation and candidate frequency point extraction: The acquired transient response signals are averaged to suppress random noise and improve the signal-to-noise ratio. A high-resolution spectral estimation method (such as a parameterized method based on an autoregressive model, or a refined FFT using a Hanning window) is used instead of the traditional simple FFT to perform a frequency domain transformation on the averaged signal. This effectively distinguishes closely adjacent spectral peaks, avoids spectral leakage, and thus more accurately locates the resonant point. All significant local maxima points are identified on the obtained spectrogram as candidate resonant frequencies.

[0028] III. Correlation between frequency point selection and physical mechanism: Candidate frequency points were screened based on their physical meaning and sensitivity to insulation defects, ultimately determining three core characteristic frequency points: Interlayer capacitance sensitive frequency ( Select the frequency point most sensitive to changes in inter-layer / inter-turn capacitance of the windings. This can be achieved by slightly altering the capacitance changes between the windings during calibration, observing which candidate frequency point exhibits the most significant shift and energy change.

[0029] Partial discharge characteristic frequency points ( Under the condition of simultaneous verification with partial discharge detection, select the frequency point where the high-frequency component energy shows a specific increase when a measurable partial discharge occurs. This frequency point should be strongly correlated with the spectral characteristics of the PD pulse.

[0030] Carbonization noise characteristic frequency points ( ): Select a frequency point that can effectively characterize broadband electromagnetic noise energy through theoretical calculations (based on winding structure dimensions) or verification on samples that have failed due to carbonization. This frequency point is usually located within a specific frequency band that can capture the radiated noise from the carbonized channel.

[0031] S2. Set the narrow bandwidth for each characteristic frequency point; Specifically, after determining three characteristic frequency points, a narrow bandwidth is defined around each frequency point. (e.g., ±5kHz).

[0032] S3. Based on the narrow bandwidth and characteristic frequency points, calculate the characteristic frequency point energy corresponding to each characteristic frequency point, and use the characteristic frequency point energy as the reference energy. The reference energy includes the first reference energy, the second reference energy, and the third reference energy.

[0033] Specifically, under the reference insulation condition, the signal energy within this narrow band is calculated to obtain the energy at each characteristic frequency point. , , The calculation method is as follows:

[0034] in Frequency point numbering, , , Original signal N-point Discrete Fourier Transform , Sampling frequency, For signal length, The reference energy is used. The original signal is an excitation signal applied to the stator windings of the motor within a predetermined frequency range.

[0035] Three preset characteristic frequencies and their energy distributions constitute a feature identification system for the correlation between the motor's insulation state and electromagnetic response. Figure 4 The extracted frequencies and their energy distributions are shown. Through multidimensional correlation analysis of frequency distribution, energy weighting, and dynamic evolution processes, this system can analyze the nonlinear evolution of dielectric parameters during insulation degradation. Revealing the degradation of dielectric properties at the molecular scale (early warning); Capture the growth and discharge activity of micron-level air gap defects (mid-term monitoring); Warning of the fatal risk of millimeter-scale carbonization channel formation (late-stage blockage).

[0036] S4. Collect multiple operating points of the motor under normal operating conditions and the expected energy corresponding to each operating point; Specifically, in actual motor operation, the speed n, load torque T, and winding temperature T are... w Operating parameters can systematically modulate the resonant energy response, creating strong interference (e.g., increased rotational speed exacerbates skin effect losses; increased load raises the common-mode voltage amplitude; temperature changes affect material conductivity and loss angle). These factors cause characteristic energy to be significantly affected even under perfectly healthy insulation conditions. , , It will also drift significantly depending on the operating conditions. Its core idea is to construct an insulated electromagnetic response twin for each online motor. This twin can predict the energy response benchmark that healthy insulation should have under any given operating condition.

[0037] S5. A baseline database is formed based on the baseline energy, operating point, and the expected energy corresponding to the operating point. Specifically, for each operating point A large number of leakage current transient signals were collected, and each characteristic frequency point was calculated. Narrowband energy mean To create a health benchmark database:

[0038] in, For rotational speed, For torque, For temperature, As the first reference energy, As the second reference energy, As the third reference energy, The desired energy.

[0039] S6. Train the energy prediction model using the benchmark database to obtain the optimal parameters, and optimize the energy prediction model based on the optimal parameters to obtain the optimal energy prediction model; The energy prediction model is trained using a benchmark database to obtain optimal parameters. The energy prediction model is then optimized based on these optimal parameters to obtain the optimal energy prediction model, which includes: Each set of training data from the benchmark database is input into the energy prediction model, and the optimal parameters are obtained by the least squares optimization algorithm. Each set of training parameters includes a working point, the benchmark energy corresponding to each feature frequency point, and the expected energy corresponding to each working point. The optimal energy prediction model is obtained by optimizing the energy prediction model based on the optimal parameters.

[0040] Specifically, using data from BaseDB, for each feature frequency point Independently fit a mathematical model. Use a quadratic polynomial model:

[0041] in, For energy prediction models, For model parameters, arrive These are the coefficients in the quadratic polynomial expansion. This is the reference energy.

[0042] This is a quadratic polynomial model with cross terms, capable of capturing the coupling effects between operating conditions. Optimal parameters are obtained using optimization algorithms such as the least squares method. This model It can predict the expected energy of healthy insulation under any operating condition. .

[0043] S7. Input the actual working conditions into the optimal energy prediction model to predict the predicted energy corresponding to each characteristic frequency point. The predicted energy includes the first predicted energy, the second predicted energy, and the third predicted energy. S8. Based on the actual working conditions and each characteristic frequency point, the first calculated energy, the second calculated energy, and the third calculated energy are calculated. Specifically, during online monitoring, the system collects current operating conditions in real time. ), input energy prediction model to obtain , , (i.e., the first predicted energy, the second predicted energy, and the third predicted energy). Simultaneously, the expected energy value for the health status under the current operating conditions is calculated. , , (That is, the first calculated energy, the second calculated energy, and the third calculated energy). The calculation methods for the first calculated energy, the second calculated energy, and the third calculated energy are the same as those for the reference energy.

[0044] S9. Based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy, calculate the first energy difference, the second energy difference, and the third energy difference. Based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy, the first energy difference, the second energy difference, and the third energy difference are calculated, including: The difference between the first predicted energy and the first calculated energy is obtained to get the first calculated difference value. The absolute value of the first calculated difference value is then obtained to get the first energy difference value. The difference between the second predicted energy and the second calculated energy is used to obtain the second calculated difference value. The absolute value of the second calculated difference value is then used to obtain the second energy difference value. The difference between the third predicted energy and the third calculated energy is used to obtain the third calculated difference value. The absolute value of the third calculated difference value is used to obtain the third energy difference value.

[0045] Specifically, the calculation methods for the first energy difference, the second energy difference, and the third energy difference are as follows:

[0046]

[0047]

[0048] in, This is the first energy difference. This is the second energy difference. This is the third energy difference.

[0049] S10. Based on the first energy difference, the second energy difference, and the third energy difference, the detection results of the permanent magnet motor stator insulation weakening are obtained.

[0050] Based on the first energy difference, the second energy difference, and the third energy difference, the results of the permanent magnet motor stator insulation weakening detection are obtained, including: Set a first threshold, a second threshold, a third threshold, a fourth threshold, and a fifth threshold; If the first energy difference is greater than the first threshold and the second and third energy differences are both less than the second threshold, then the result of the permanent magnet motor stator insulation weakening detection is early deterioration. If the second energy difference is greater than the third threshold, the third energy difference is less than the fourth threshold, and the first energy difference continues to rise, then the result of the permanent magnet motor stator insulation weakening test is mid-term deterioration. If the third energy difference is greater than the fifth threshold or the rate of change of adjacent third energy differences is greater than the preset rate of change, then the result of the permanent magnet motor stator insulation weakening detection is late-stage deterioration.

[0051] Specifically, this scheme's tiered early warning mechanism abandons the traditional single-threshold criterion, not only judging the current state but also assessing the deterioration trend and risk confidence level, thus achieving a leap from alarm to predictive diagnosis. Early warning decisions do not rely on the three ΔE indicators in isolation but comprehensively consider their combination patterns and evolutionary weights to form a joint diagnostic conclusion. Its core decision-making logic is as follows: Figure 5 As shown.

[0052] Early deterioration: >20% (first threshold), and , No significant change (<20%) (second threshold).

[0053] Physical significance: At this stage, the interlayer distributed capacitance has shown an observable increase (>5%), mainly caused by the degradation of dielectric properties due to the breakage of the insulating material molecular chains caused by thermal and electrical stress. The system triggers an "early warning," indicating the need to increase the density of operational monitoring.

[0054] Medium term deterioration: >40% (third threshold), and It continues to rise or remains at a high level. It remains at a low level (<30%) (fourth threshold).

[0055] Physical significance: This indicates an increase in micron-sized air gaps or cracks within the insulation, leading to more active and intensified partial discharge activity, a clear sign of accelerated insulation aging. The system has triggered a "mid-term warning," and inspection and maintenance should be planned for the near future.

[0056] Failure warning: >80% (fifth threshold), or observed Rapid, step-like growth occurs (e.g., growth exceeding 20% ​​within 24 hours (preset rate of change)).

[0057] Physical significance: This indicates that localized carbonization has occurred in the insulation, forming millimeter-scale conductive channels. Ohmic losses have increased dramatically, and a through-circuit may occur at any time. The system immediately triggers an "emergency warning" and requires immediate shutdown for maintenance to prevent catastrophic failure.

[0058] This dynamic compensation architecture essentially constructs an insulated electromagnetic digital twin for each online-running motor. The compensated energy offset... It can be regarded as an intrinsic state parameter characterizing the degree of insulation degradation, and a decision-making mechanism based on multi-feature collaboration enables in-depth interpretation and intelligent adjudication of this parameter. It is not affected by the motor operating mode, but only reflects the degradation process of the electrochemical and physical structure of the insulation material itself, providing a highly reliable decision-making basis with action-guiding value for the transformation from "periodic maintenance" to "predictive maintenance".

[0059] A permanent magnet motor stator insulation weakening detection system based on leakage current, comprising: A calibration module is used to calibrate the characteristic frequency points of the permanent magnet motor, wherein there are three characteristic frequency points; The setting module is used to set the narrow bandwidth for each of the characteristic frequency points; The first calculation module is used to calculate the characteristic frequency energy corresponding to each characteristic frequency point based on the narrow bandwidth and characteristic frequency points, and use the characteristic frequency energy as the reference energy. The data acquisition module is used to collect multiple operating points of the motor when it is operating within the normal operating range, as well as the expected energy corresponding to each operating point. The database construction module is used to form a benchmark database based on the benchmark energy, the operating point, and the expected energy corresponding to the operating point. The optimization module is used to train the energy prediction model using the benchmark database, obtain the optimal parameters, and optimize the energy prediction model based on the optimal parameters to obtain the optimal energy prediction model. The second calculation module is used to input the actual working conditions into the optimal energy prediction model and predict the predicted energy corresponding to each characteristic frequency point. The predicted energy includes the first predicted energy, the second predicted energy, and the third predicted energy. The third calculation module is used to calculate the first calculation energy, the second calculation energy, and the third calculation energy based on the actual working conditions and each characteristic frequency point. The fourth calculation module is used to calculate the first energy difference, the second energy difference, and the third energy difference based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy. The detection module is used to obtain the detection result of the weakening of the stator insulation of the permanent magnet motor based on the first energy difference, the second energy difference, and the third energy difference.

[0060] This application also discloses a terminal device, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor loads and executes the computer program, it employs a method for detecting the weakening of the stator insulation of a permanent magnet motor based on leakage current.

[0061] The terminal device can be a computer device such as a desktop computer, a laptop computer, or a cloud server. The terminal device includes, but is not limited to, a processor and a memory. For example, the terminal device may also include input / output devices, network access devices, and buses.

[0062] The processor can be a central processing unit (CPU). Of course, depending on the actual use, it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc., and this application does not limit it.

[0063] The memory can be an internal storage unit of the terminal device, such as a hard disk or RAM of the terminal device, or an external storage device of the terminal device, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD), or flash memory card (FC) equipped on the terminal device. Furthermore, the memory can be a combination of internal storage units and external storage devices of the terminal device. The memory is used to store computer programs and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or will be output. This application does not limit this.

[0064] In this terminal device, a method for detecting the weakening of permanent magnet motor stator insulation based on leakage current in the above embodiments is stored in the memory of the terminal device and loaded and executed on the processor of the terminal device for convenient use.

[0065] This application also discloses a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it employs a method for detecting weakened stator insulation of a permanent magnet motor based on leakage current, as described in the above embodiments.

[0066] The computer program can be stored in a computer-readable medium. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or certain middleware. The computer-readable medium includes any entity or device capable of carrying computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the computer-readable medium includes, but is not limited to, the above-mentioned components.

[0067] The above-described method for detecting weakened stator insulation of a permanent magnet motor based on leakage current is stored in the computer-readable storage medium and loaded and executed on the processor to facilitate the storage and application of the above method.

[0068] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of protection of this application is limited to these examples; under the concept of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of one or more embodiments of this application as described above, which are not provided in detail for the sake of brevity.

[0069] One or more embodiments in this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this application. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of one or more embodiments in this application should be included within the protection scope of this application.

Claims

1. A method for detecting stator insulation weakening in permanent magnet motors based on leakage current, characterized in that, include: The characteristic frequency points of the permanent magnet motor are calibrated, and there are 3 characteristic frequency points; Set a narrow bandwidth for each of the aforementioned characteristic frequency points; Based on the narrow bandwidth and the characteristic frequency points, the characteristic frequency point energy corresponding to each characteristic frequency point is calculated, and the characteristic frequency point energy is used as the reference energy. Collect multiple operating points of the permanent magnet motor under normal operating conditions and the expected energy corresponding to each operating point; A benchmark database is formed based on the benchmark energy, operating point, and the expected energy corresponding to the operating point. The energy prediction model is trained using the benchmark database to obtain the optimal parameters, and the energy prediction model is optimized based on the optimal parameters to obtain the optimal energy prediction model. The actual working conditions are input into the optimal energy prediction model to predict the predicted energy corresponding to each characteristic frequency point. The predicted energy includes the first predicted energy, the second predicted energy, and the third predicted energy. Based on the actual working conditions and each characteristic frequency point, the first calculated energy, the second calculated energy, and the third calculated energy are calculated. Based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy, the first energy difference, the second energy difference, and the third energy difference are calculated. Based on the first energy difference, the second energy difference, and the third energy difference, the detection results of the permanent magnet motor stator insulation weakening are obtained.

2. The method as described in claim 1, characterized in that, The step of calculating the characteristic frequency energy corresponding to each characteristic frequency point based on the narrow bandwidth and characteristic frequency points, and using the characteristic frequency energy as the reference energy, includes: Set the sampling frequency; Based on the narrow bandwidth and the characteristic frequency point, the upper cutoff frequency and the lower cutoff frequency are obtained; Obtain the original signal when calibrating the characteristic frequency point of the permanent magnet motor; Based on the original signal, the upper cutoff frequency, the lower cutoff frequency, and the sampling frequency, the characteristic frequency energy corresponding to each characteristic frequency point is calculated, and the characteristic frequency energy is used as the reference energy.

3. The method as described in claim 1, characterized in that, Each operating point includes speed, torque, and temperature.

4. The method as described in claim 1, characterized in that, The process of training an energy prediction model using the benchmark database to obtain optimal parameters, and then optimizing the energy prediction model based on the optimal parameters to obtain the optimal energy prediction model, includes: Each set of training data in the benchmark database is input into the energy prediction model, and the optimal parameters are obtained by the least squares optimization algorithm. Each set of training parameters includes a working point, the benchmark energy corresponding to each feature frequency point, and the expected energy corresponding to each working point. The energy prediction model is optimized based on the optimal parameters to obtain the optimal energy prediction model.

5. The method as described in claim 1, characterized in that, The step of calculating the first energy difference, the second energy difference, and the third energy difference based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy includes: The difference between the first predicted energy and the first calculated energy is obtained to get the first calculated difference value. The absolute value of the first calculated difference value is then obtained to get the first energy difference value. The difference between the second predicted energy and the second calculated energy is obtained to get the second calculated difference value. The absolute value of the second calculated difference value is then obtained to get the second energy difference value. The difference between the third predicted energy and the third calculated energy is used to obtain the third calculated difference value. The absolute value of the third calculated difference value is then used to obtain the third energy difference value.

6. The method as described in claim 1, characterized in that, The step of obtaining the permanent magnet motor stator insulation weakening detection result based on the first energy difference, the second energy difference, and the third energy difference includes: Set a first threshold, a second threshold, a third threshold, a fourth threshold, and a fifth threshold; If the first energy difference is greater than the first threshold and the second and third energy differences are both less than the second threshold, then the result of the permanent magnet motor stator insulation weakening detection is early deterioration. If the second energy difference is greater than the third threshold, the third energy difference is less than the fourth threshold, and the first energy difference continues to rise, then the result of the permanent magnet motor stator insulation weakening test is mid-term deterioration. If the third energy difference is greater than the fifth threshold or the rate of change of adjacent third energy differences is greater than the preset rate of change, then the result of the permanent magnet motor stator insulation weakening detection is late-stage deterioration.

7. The method of claim 6, characterized in that, The first threshold is less than the third threshold, and the third threshold is less than the fifth threshold.

8. A permanent magnet motor stator insulation weakening detection system based on leakage current, characterized in that, include: A calibration module is used to calibrate the characteristic frequency points of the permanent magnet motor, wherein there are three characteristic frequency points; The setting module is used to set the narrow bandwidth for each of the characteristic frequency points; The first calculation module is used to calculate the characteristic frequency energy corresponding to each characteristic frequency point based on the narrow bandwidth and characteristic frequency points, and use the characteristic frequency energy as the reference energy. The data acquisition module is used to collect multiple operating points of the motor when it is operating within the normal operating range, as well as the expected energy corresponding to each operating point. The database construction module is used to form a benchmark database based on the benchmark energy, the operating point, and the expected energy corresponding to the operating point. The optimization module is used to train the energy prediction model using the benchmark database, obtain the optimal parameters, and optimize the energy prediction model based on the optimal parameters to obtain the optimal energy prediction model. The second calculation module is used to input the actual working conditions into the optimal energy prediction model and predict the predicted energy corresponding to each characteristic frequency point. The predicted energy includes the first predicted energy, the second predicted energy, and the third predicted energy. The third calculation module is used to calculate the first calculation energy, the second calculation energy, and the third calculation energy based on the actual working conditions and each characteristic frequency point. The fourth calculation module is used to calculate the first energy difference, the second energy difference, and the third energy difference based on the first predicted energy, the second predicted energy, the third predicted energy, the first calculated energy, the second calculated energy, and the third calculated energy. The detection module is used to obtain the detection result of the weakening of the stator insulation of the permanent magnet motor based on the first energy difference, the second energy difference, and the third energy difference.

9. A terminal device, comprising a memory and a processor, characterized in that, The memory stores a computer program that can run on a processor, and when the processor loads and executes the computer program, it employs the method described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by the processor, it employs the method described in any one of claims 1 to 7.

Images